Signature stats
Turnover Ledger: takeaways and giveaways valued by zone
Giveaway and takeaway totals are among the worst stats in the box score. They depend on who is scoring the game, they count a harmless neutral-zone turnover the same as a cough-up in the slot, and their meaning changed in 2024-25. The Turnover Ledger fixes the second problem and flags the other two.
Updated · Method version sequences v1
How it's calculated
Ledger = Σ_zones TK/60(z) × V_TK(z) − Σ_zones GV/60(z) × V_GV(opp zone)
- TK/60(z), GV/60(z)
- the player's takeaways and giveaways per 60 in each zone
- V
- league xG the gaining team generates within 20 s, cut at a whistle, by the zone where it gains the puck
- opp zone
- a giveaway in your own zone gives the opponent the puck in its offensive zone
| Turnover | Gaining team gets the puck in | xG in 20 s |
|---|---|---|
| Takeaway | Offensive zone | 0.030 |
| Opponent giveaway | Offensive zone | 0.029 |
| Takeaway | Neutral zone | 0.019 |
| Takeaway | Defensive zone | 0.016 |
| Opponent giveaway | Neutral zone | 0.014 |
| Opponent giveaway | Defensive zone | 0.014 |
The taker gets the credit and the giver gets the cost. We publish rates times league values rather than what actually happened after each turnover: whether a particular giveaway led to a goal is mostly luck, but how often a player turns the puck over in dangerous places is repeatable.
What a turnover is worth, by where the puck is won
Research sample: 200 games from 2025-26 (2025020500–699)
xG for the gaining team in the next 20 s ranges from 0.030 for an offensive-zone takeaway to 0.014 for an opponent giveaway in the neutral or defensive zone.
Worked example
A forechecking winger
Includes example dataPer 60 minutes at 5v5, a winger has 1.2 offensive-zone, 0.6 neutral-zone and 0.4 defensive-zone takeaways. He gives the puck away 0.8 times in his own zone, 0.7 in the neutral zone and 0.9 in the offensive zone.
- Credit: 1.2 × 0.030 + 0.6 × 0.019 + 0.4 × 0.016 = 0.036 + 0.0114 + 0.0064 = 0.0538 xG/60.
- Cost: own-zone giveaways hand the opponent the puck in its offensive zone, 0.8 × 0.029 = 0.0232. Neutral zone: 0.7 × 0.014 = 0.0098. Offensive zone (opponent gains in its own end): 0.9 × 0.014 = 0.0126. Total 0.0456 xG/60.
- Net: 0.0538 − 0.0456 = +0.008 xG/60.
Slightly positive: his offensive-zone takeaways more than pay for his giveaways, even though he has more giveaways (2.4/60) than takeaways (2.2/60). The rates are illustrative; the zone values are the league's.
The 2024-25 recording change
Turnovers are logged by off-ice scorers in each arena. From 2017-18 to 2023-24, arenas disagreed wildly: one rink's scorers logged 0.36 times as many giveaways as the same teams were credited with on the road, and another 1.74 times, and these rink factors were persistent (half-season correlations of 0.71–0.86). From 2024-25 the pattern changed across the league at once.
| Legacy (2017-18 to 2023-24) | Centralised (2024-25 on) | |
|---|---|---|
| Giveaways per game (league) | 14.6 | 29.4 |
| Takeaways per game (league) | 14.0 | 9.6 |
| Arena giveaway factor range | 0.36× to 1.74× | Noise |
| Arena factor half-season r | 0.71 to 0.86 | Noise |
| Arena hit factor range | 0.63× to 1.46× | Noise |
- Every turnover and hit row carries a recording regime flag: legacy (2023-24 and earlier) or centralised (2024-25 on).
- Legacy seasons get a shrunk arena adjustment: factor = home-arena rate ÷ same teams' road rate, shrunk toward 1 by its reliability. Centralised seasons are not adjusted.
- Leaderboards use per-season percentiles, never raw counts across the break, and legacy turnover and hit stats show a "legacy recording" badge.
Reliability
Turnover rates by zone are repeatable; what happens after any single turnover is not. That is why the ledger uses league values.
| Measure | Split-half r | Sample | Verdict |
|---|---|---|---|
| Per-event outcome (xG after a player's turnovers)Shown only as a "what happened" view. | ≈ 0.1 | Split-half, 200 games | Noise |
| Turnover rate by zoneThe rate is stable; exact r not yet published. | — | Half a season | Moderate |
| Legacy arena factorsStable bias, which is why we adjust for it. | 0.71–0.86 | Half seasons, 2017-18 to 2023-24 | Stable |
Split-half r: we split each sample in two (for example odd and even games) and correlate the two halves across players or teams. Near 0 means the stat is mostly noise; near 1 means it is a stable trait.
Limitations
- Takeaways and giveaways are judgement calls by the scorer. A giveaway on one night can be a takeaway for the other team on another.
- Turnovers without coordinates (common in older seasons) cannot be assigned a zone and are left out of the ledger.
- The 20 s window and whistle cut-off miss slower consequences, such as a turnover that leads to an icing and a tired defensive-zone draw.
- Legacy-season adjustments reduce arena bias but can't remove it. Treat pre-2024-25 turnover numbers as rough.
FAQ
Why did NHL giveaways double in 2024-25?
Scorer recording changed across the league. League giveaways per game went from 14.6 to 29.4 and takeaways fell from 14.0 to 9.6, while the large arena-to-arena differences of earlier seasons disappeared. Players did not suddenly get careless.
Are giveaways bad?
It depends where. An opponent giveaway in its own zone is worth 0.029 xG to the team that gets the puck; one near the other team's blue line is worth 0.014. Players who handle the puck a lot also give it away a lot, so raw totals punish puck carriers.
Why not use what actually happened after each turnover?
Because it is noise. Whether a given turnover led to a shot or goal had a split-half correlation of about 0.1 per player. The rate of turnovers by zone is stable, so we multiply it by league values.
Can I compare 2022-23 and 2025-26 turnover stats?
Not as raw counts. Use per-season percentiles, which is what our leaderboards show. Legacy seasons are marked with a badge.